JOHN D.HOFFMAN and ROBERT L.MILLER

Compared to metallurgy and ceramics, the field of organic polymers is new. For example, the now widely used polymer polyethylene was discovered in Britain only a little more than 50 years ago. New polymers continue to be introduced every year at a significant rate, and new applications in both science and commerce inevitably follow. The number of new polymers is sufficient to justify books on the subject. ¹

This review considers some of the opportunities and challenges related to organic polymers. It not only covers what polymers are but also develops the nature of certain opportunities. Chemical properties and physical properties (which are becoming very important) are illustrated, both in experiment and in theory. Our goal in part is to remove some of the mystique that sometimes surrounds the subject of organic polymers and polymer science and to put these substances in proper perspective in the larger world of materials generally. The need for understanding the processing of polymers is one of the main themes of the review, as is the nature of certain exciting new theoretical developments.

Roles of the molecular theories and of fundamental research are broadly illustrated throughout. The very strong interaction of polymers with other materials and their significant contribution to application technology in other fields are shown. To balance this, instances are mentioned where polymer science and technology owe a debt to other areas of materials research and development, and examples are given where polymers indeed have their limitations. Although this review covers organic polymers, it does not intrude far into the field of biotechnology, which would require a separate review of its own. Emphasis is given to major new trends, such as the use of polymers

in composites, and the special properties that can be achieved from polymer blends.

Figure 1 shows one aspect of the sometimes unique behavior of polymers. It depicts a typical long, concatenated polymer chain adsorbed on a surface. The chain consists of similar chain units (monomers) and is flexible. This illustration is important because it shows that, to desorb the entire polymer molecule, all of the little “feet” in the long molecule that are attached to points on the surface must be lifted from the surface simultaneously. From a probability point of view this is difficult even when the occasional attachments involve relatively weak van der Waals forces, and it is still more difficult if the attachments involve chemical bonds. The figure thus indicates one reason that polymers make such good surface coatings: Once adsorbed, they can be very difficult to detach. Polymer molecules will adsorb on a surface in seconds but may take weeks to detach fully, even in the presence of pure solvent; in poor solvents or in the absence of solvents, they virtually never detach. For short chains this is not the case; from a statistical point of view they come off rapidly in solvents because of the relatively low number of attachments. This simple illustration shows that the treatment of polymers deals with a property not ordinarily thought of as an important materials parameter, namely, molecular length, or, in more customary terms, molecular weight.

In the field of organic polymers, a wide range of chemical structures is readily available. As a beginning, consider the remarkable variety of properties and morphologies one can obtain with a specific single polymer chain, i.e., with “constant chemistry.” For this purpose we emphasize polyethylene,—(CH₂—CH₂) _n —, which is a very simple chain. The examples will be single crystals, lamellar spherulitic structures, and high-strength fibers—all with the same molecule (common polyethylene), but with different processing.

Consider the following experiment: in ordinary xylene at, say, 135 to 138°C, a small amount of linear polyethylene (0.001 to 0.01 percent) is dissolved, and the solution is cooled to around 70 to 80°C. Chains with a

molecular weight of 50,000 (corresponding to a chain length of about 455 nm) are suitable. Crystals, such as the one shown in Figure 2, will form and precipitate. This is a polyethylene single crystal! That such beautiful crystals could be formed from a polymer came as a surprise to most researchers. In now classical work, Keller ² elucidated the basic nature of these crystals. Figure 2 is an electron micrograph of such a crystal taken by Bassett, ³ formerly one of Keller’s students. Although a full discussion of polymer single crystals is beyond the scope of this review, salient features are presented. A somewhat idealized structure of such a crystal ⁴ is shown in Figure 3, in which each continuous, accordion-like line represents a single long polyethylene chain. It has been well established that the polymer molecules are chain-folded as shown. The diagonal striations seen in Figure 2 are the (310) slip planes indicated in Figure 3. The thickness l of a crystal is, say, 9 to 20 nm and is dependent on the crystallization temperature, as suggested by nucleation theory. ⁴ (Nucleation theory, invented by metallurgists, is highly useful in explaining the formation and thickness of these polymer single crystals.)

Such crystals began a revolution in polymer physics, namely, as a consequence of the chain-folding phenomenon. Modern techniques such as infrared spectroscopy ⁵ and neutron scattering ^{6 , 7} suggest that the fold perfection in such crystals is about 75 percent—nature does in fact make mistakes in putting together a crystal consisting of such long molecules. (The concept

of chain folding—even in single crystals—was adamantly resisted by some of the doyens in the field. In the end, the concept proved too useful and well supported to be ignored.)

Polymer single crystals are not of great commercial importance, although strong mats resembling a sheet of paper can be prepared from them. Their importance is that they started a whole new line of thought concerning the morphology and basic character of crystalline polymers and, moreover, provided new insights into structure-property relationships for crystalline and semicrystalline polymeric materials. Many (but not all) of the polymers of commerce are potentially crystallizable, and in practice do frequently exhibit crystalline, or more often semicrystalline, properties. Depending on intended use, the crystallinity can be either useful or detrimental. More frequently it is useful. For example, the crystalline regions are relatively effective barriers to diffusion of gases and small molecules. Hence the use of semicrystalline polymers—e.g., polyethylene—in applications such as food wrapping. Also, the crystalline regions may act as physical, rather than chemical, cross-links in forming a three-dimensional network that imparts mechanical stability. Yet, when melted again they may be processed or even reprocessed.

Ordinary commercially crystallized polyethylene, such as food and freezer wrapping material, is crystallized from the melt. Under such conditions, objects such as those shown in Figure 4 are frequently seen under a polarizing optical microscope. ⁸ These are called spherulites (by analogy with mineralogical spherulites) and appear superficially to be exceedingly different from a single crystal such as that shown in Figure 2. The four fields in Figure 4 represent three different molecular weights of polyethylene (18,000; 30,000; and 60,000). The first three were crystallized isothermally at the temperature indicated; the fourth was quench-crystallized. Molecular weights studied varied from 3,600 to 807,000, that is, chains varying from 32 nm to 7,300 nm in length. Commercial polyethylene crystallized in an unstrained manner contains typical spherulites as do other crystallizable polymers, such as nylon, which is a polyamide.

What is the structure of a spherulite and what is the relation, if any, with single crystals such as that in Figure 2? Figure 5 shows some of the structural details of a polymer spherulite. It consists of lamellae or blades radiating from a central point, which is usually a piece of dirt (more elegantly, a “heterogeneous nucleus”). The lamellae are again chain-folded, somewhat like a single crystal, although not as perfectly. In spherulites, however, there are interlamellar links that make it stronger, and there are branch points that allow the spherulite to be a three-dimensional object. Nevertheless, the basic structural unit is similar, but not exactly equal, to that of the single crystal.

The lamellar nature of polymer spherulites in melt-crystallized polymer was first recognized by Eppe, Fischer, and Stuart ⁹ in Germany shortly after Keller’s original work in England ² on single crystals from dilute solution. The bands seen in Figure 4D arise from the cooperative twisting of adjacent lamellae. ¹⁰ This lamellar nature is more clearly seen in Figure 6, ¹¹ which is an electron micrograph of a microtomed section of a spherulite of polyethylene. The lamellae are being viewed edge-on and have the appearance of the edges of a fanned deck of cards. ¹² The lamellae in Figure 6 are about 30 nm thick, and the polyethylene sample is a good polymer fraction with about the same molecular weight as the single crystal of Figure 2. (A “fraction” is a polymer specimen for which special techniques have been employed to ensure that the polymer chains are about the same length; most polymers, as synthesized, have a broad distribution of lengths.) Impurities and the shorter polymer chains are normally excluded from such lamellae. ¹³ Some of the shorter chains may subsequently crystallize; the remainder of the

excluded material contributes to the noncrystalline (amorphous) component of the spherulite.

Spherulites from branched polyethylene, i.e., polyethylene with adventitious pendant—CH₃ and—CH₂CH₃ groups, do not display as clear a lamellar picture. Polyethylene spherulites, then, present an interesting morphology in which 15 to 20 percent of the material in the objects in Figure 4 is amorphous and the remainder is like that of a single crystal. The amorphous material is largely between the lamellae, and part of it is in the form of interlamellar links. Since composites are discussed later, note that the polymer spherulite is a self-assembled composite. That is one of the reasons polymers such as polyethylene and nylon are useful. They are natural composites, held together in part by interlamellar links, the reinforcement coming from the crystal lamellae. The composite is formed by the process of crystallization itself, an interesting result. Self-assembly is mentioned again below.

Another mode of solution crystallization is possible while still holding the chemistry constant. Solutions, such as those used to prepare polyethylene single crystals, can be crystallized under shear. Vigorous stirring is sufficient, and this process works best if the chains are long. Here a high molecular weight is desirable, and a more concentrated solution (approximately 0.1 to 1.0 percent) is to be used. Under these conditions a strikingly different morphological entity is obtained, as shown in Figure 7 (lying on a graphite substrate). ¹⁴ The central thread is a very strong fiber. Work of this nature was first performed by Pennings and Kiel ¹⁵ in Holland (which highlights the international character of the significant advances in this field). A break in the central thread can be seen roughly in the middle of the lower unit—it was stressed too much while being subjected to the beam in an electron microscope. The shear direction during crystallization was parallel to the central threads. Such entities are called “shish kebabs” and display the structure shown in Figure 8 ¹⁶ —a tremendously strong central fiber of primarily extended-chain conformation decorated with (again) chain-folded lamellae. The chain-folding part is not as perfect as it is in a single crystal, but the extended-chain perfection in the central fiber is high. The important point here is that shish kebabs are enormously strong—at least half the strength of the carbon-carbon bond and, in relation to their weight, stronger

than steel. The strength, of course, is due to the “shish.” Although shish kebabs themselves are not commercially important, they have been the impetus behind worldwide efforts to produce fibers, sheets, and rods commercially whose strengths take advantage of the molecular orientation of the shish kebab structure. This area of processing is known to polymer scientists as stress-induced crystallization (SIC).

Briefly, one the approaches used to exploit the shish kebab effect is that of solid-state extrusion of polymers in the absence of solvents (for a recent review, see note 17). Figure 9 shows the results for polyethylene. ¹⁸ The thick, opaque rod at the upper right of this figure consists of common, spherulitic, melt-crystallized polyethylene, as discussed earlier. The extrusion process breaks up the spherulitic structure and produces optically clear fibers, as depicted in the lower portion of the figure. Such processes are being commercialized in laboratories in many countries, for example, by the Allied Corporation in the United States. The structure of such extruded polymers proposed by Zachariades and Porter ¹⁹ is shown schematically in Figure 10 (others have presented similar pictures). The resemblance to the

structure of the shish kebab in Figure 8 is clear—mostly concatenated chains in the core fibril (the “shish”) with perhaps some residual folds. Whatever the fine details, the important point is that the structure consists primarily of long molecules that are parallel over long distances. Only in that way can the enormous tensile strength and high modulus of such materials be explained.

A considerable amount of the progress cited here depended on electron microscopy. Polymer science has made its definite contributions, but it owes a considerable debt to those scientists and engineers who made the electron microscope a practical laboratory instrument.

Table 1 lists the mechanical properties (modulus) of several common materials (the modulus of a material is the initial slope of its stress-strain curve). The first material is a soft metal (aluminum), followed by glass fibers, and so on, ending with polyethylene in different forms. Moduli vary from 70 to 420 GPa, with that of extruded polyethylene fibers being essentially equal to that of steel. Theoretically, the modulus of polyethylene should be about 300 GPa; experimentally, the best achieved to date is about 220 GPa. Also shown in Table 1 is the specific modulus, which is the modulus divided by the density. On a weight basis, then, the polymeric fibers are impressive,

and one can understand why polyethylene, an inexpensive raw material, should be the subject of such interest. However, there is a negative aspect to this product. Although the longitudinal strength is high, the strength in the transverse direction is not spectacular; such materials tend to fibrillate readily.

Thus, with a single type of polymer chain, three different morphological structures can be prepared (with vastly differing mechanical properties), depending on the processing variables chosen. This story, with variations, is repeated for many crystallizable polymers, such as the polyamides (nylons) and the polyesters.

Polymers do have their problems, not the least of which is instability at high temperatures. Here, they do not challenge metals or ceramics, although some are amazingly stable. When rendered highly thermally stable by the introduction of suitable chemical chain units, they tend to be more difficult to process. But, in many applications, often in conjunction with metals or ceramics, they are the material of choice. An additional problem with polymers is connected with their disposal as waste. However, many objects made from them can be either reused or recycled, but “mixtures” of them are hard to deal with and sorting for recycling is a definite problem.

The structure-processing-property theme may be expanded by changing the chemistry. Consider the case of polypropylene,—[CH₂—CH(CH₃)] _n —, which is “polyethylene” with a pendant methyl group on every other carbon atom. Solid-state extrusion in the manner described earlier produces uniaxially oriented material analogous to that shown in Figure 9, with again a presumably similar molecular picture. Alternatively, somewhat different processing yields biaxial orientation, which provides a superior product for certain applications. That this is not trivial is demonstrated by tests in which two polypropylene plates, each 3 mm thick, have stopped a .38-caliber normally loaded revolver bullet. This is a convincing demonstration of the impact strength possibilities of polymers; their ability to absorb large quantities of energy under certain specified conditions can be spectacular.

This discussion can be summarized as follows: (1) processing variables have a remarkable effect on the properties of polymers, even when the polymer chemistry is the same; (2) changes in the chemistry provide a greatly enhanced range of properties and opportunities. These are important themes whose initial impetus was largely from basic scientific studies, much of which started in Europe. Keller and his discovery of chain folding in single crystals ² and Pennings and Kiel and their beautiful experiments on oriented fibers ¹⁵ are examples. To these must be added the role of distinguished chemists who discovered new synthesis routes to more uniform chain structures, such as Ziegler of Germany and Natta of Italy (discussed in the next section). Clearly, fundamental research in this field has contributed to applied technology, and vice versa.

If the emphasis in the foregoing discussion seems to be too much on polyethylene, it is worth observing that polyethylene is the largest-selling polymer from both poundage and monetary standpoints. The 1985 yearly production was roughly 7 million metric tons. The profits (or losses) were mostly in dollars, pounds, yen, marks, francs, lira, or guilders.

The story does not end with crystalline polymers. There are amorphous polymers of great importance that must be considered briefly and that introduce phenomena not encountered when discussing ordinary molecules. One such phenomenon is tacticity, a certain detail of the chain microstructure. Another is a fundamental physical phenomenon called the glass transition, which is still being studied and is exceptionally important in both physical and chemical properties. A third phenomenon concerns new insight into diffusion and into how polymer molecules move. The mode of motion of a long, snake-like molecule can differ from the diffusion of, say, an atom in a metallic alloy.

The simplest chemical structure of polypropylene was indicated earlier; it is essentially polyethylene with a methyl group attached to every other carbon atom. That makes every other carbon atom in the chain chiral (i.e., asymmetric), with the result that there are three classes of polypropylene structure, depending on the chiral relationship of each pendant methyl group with its neighbors. These three forms are illustrated in the lower part of Figure 11. When all chiral atoms have the same chirality, the polymer is called isotactic,

is crystalline, and has a moderately high melting point. When the chirality alternates regularly, as in the next line of the figure, the polymer is called syndiotactic, is also crystalline, and has a moderate melting point. When the chirality of successive atoms is random, as in the last line of the figure, the polymer is called atactic and it is noncrystalline (amorphous). That is, the stereoregular polymers crystallize and the stereoirregular ones do not. The whole subject of stereoregularity in polymers was elucidated by Ziegler et al. ²¹ and Natta et al. ²² Regularity of structure is a requirement for crystallization of polymers, as indicated also in the upper part of Figure 11, in which the regular structure of polyethylene is compared with the irregular structure produced by randomly coupling two different vinyl-type monomers. The ability or inability to crystallize is used to illustrate the importance that details of chemical structure (the microstructure) have on resulting physical properties. Chemically regular chains may crystallize; irregular chains cannot. This is an additional chain property that has expanded the range of opportunities for the use of organic polymers.

In amorphous polymers such as ordinary (atactic) polystyrene, an all-pervasive phenomenon of great importance occurs wherein properties change notably over a small temperature range. A change in slope in a volume-temperature plot (also in an enthalpy-temperature plot) occurs at a moderately well-defined temperature, known as the glass transition temperature, T _g , as illustrated in the left-hand side of Figure 12. This is formally a second-order transition in the Ehrenfest sense, although nonequilibrium rate effects attend it on all real occasions. Although this change of slope in itself appears to be subtle, there is a large change in mechanical properties as a sample goes through the glass transition. For example, the mechanical modulus decreases dramatically above the glass transition temperature, as indicated in the righthand side of Figure 12: Below this temperature the material is brittle; above it, it is rubbery. (Remarkable changes in other properties also take place near T _g , such as in the dielectric loss.) This behavior also occurs in the amorphous regions of semicrystalline polymers.

As indicated earlier, the usual melt-crystallized specimen of polyethylene is partly crystalline and partly amorphous. Consequently, polyethylene exhibits amorphous properties (such as the glass transition) from the interlamellar regions, crystalline properties from the lamellae, and composite properties from the self-assembled combination. It is important to know about the glass transition, which should be thought of in the practical sense as an engineering property. Modulus behavior of a polymer is a major factor determining its behavior in use. Changes in modulus can be dramatic; frequently the modulus

decreases by orders of magnitude above the glass transition temperature. Such behavior is also exhibited by some simple molecules and inorganic glasses, but the glass transition is of exceptional importance in polymeric materials.

The structure of amorphous polymers is often represented as entangled threads ²³ according to Flory’s ²⁴ concept of polymer chains in the liquid being entangled random coils. Some polymer chains are somewhat stiffer than others, and some co-alignment of chain segments can occur in these (stiff, rod-like polymer chains can form “liquid crystals”). This picture may in some respects be approximate, but it is by far the best depiction currently available, especially for highly flexible chains. Such a picture raises two questions: (1) How does such a molecule move in such an entangled situation? and (2) What is the nature of the glass transition that appears in such systems at sufficiently low temperatures?

Figure 13 shows one way of looking at the first question. ²⁵ The answer is that the polymer molecule moves by a process of curvilinear diffusion, called reptation. This is a new concept, due to de Gennes ²⁶ and to Doi and Edwards, ²⁷ that is appearing more and more in considerations of polymer dynamics and that is extremely important. In essence, the long molecule creeps along lengthwise, thus moving its center of mass in an effective manner.

Consider the uppermost molecule shown in Figure 13; surrounding molecules are not shown. It can be characterized by a mean-square end-to-end distance, 〈R ²〉, a mean-square contour length, 〈L ²〉, and a center-of-mass, CM. Consider that after a time the molecule has moved to the lower position, in which the center of mass has moved to the second position. There are two ways this can be treated, as indicated in the figure. One (on the right) is to consider the center of mass diffusion, which is related to the ordinary diffusion coefficient of the molecule as a whole and is given by the relationship shown in the upper right, where D _cm is the diffusion coefficient of the change of center of mass. This is a well-known formula by which the mean-square end-to-end distance, 〈R ²〉, is related to the time, t _r , for the center of mass to move a distance equal to 〈R ²〉. The time is defined as shown, and D _cm can be obtained by ordinary diffusion techniques.

Alternatively, this problem can be considered from the standpoint of reptation (left-hand side). Here, the molecule moves by lengthwise translation. If the diffusion coefficient is defined in units of lengthwise motion, the time, t _r (which is the time to move one contour length), can be written in terms of the mean-square contour length, 〈L ²〉, as shown. However, the two times are equal, and the diffusion coefficient for curvilinear motion, D _r , can be expressed as shown in the first box. This quantity can be determined since all quantities on the right-hand side are measurable. This is another of de Gennes’ results, ²⁶ as adapted in Figure 13 by DiMarzio, ²⁵ and yields values

of the curvilinear diffusion coefficient for reptation, D _r . For most organic polymers, 〈L ²〉>>〈R ² 〉 with the result that reptational diffusion will be greatly enhanced relative to ordinary diffusion. Reptation is an effective mode of moving the center of mass of a polymer molecule.

One purpose of discussing motion in terms of reptation is to be able to explain how the polymer crystal discussed earlier could be formed out of an entangled melt. The answer is indicated schematically in Figure 14, in which the not-yet-crystallized portion of the polymer molecule is shown within a “reptation tube” from which the molecule is “reeled in” by the force of crystallization and forms chain folds on the crystal surface. ²⁸ The concept of reptation is absolutely necessary to understand how a polymer chain can

crystallize out of an entangled melt within a reasonable time. No special “pre-structures” in the liquid are required to permit crystallization-from the melt. Naturally, the longer the reptation tube (that is, the longer the molecule), the more difficult is the reeling-in ²⁹ process. Figure 14 explains pictorially why crystallization becomes more difficult at higher molecular weights. Over and beyond its application to crystallization, reptation has a significant potential impact that is discussed below.

The origin and nature of the glass transition in amorphous substances in general and amorphous polymers in particular have been the subject of much study. In the polymer case, Gibbs and DiMarzio showed in a statistical mechanical treatment ³⁰ that it is highly plausible that a true second-order thermodynamic transition underlies it. This thermodynamic transition temperature, denoted T ₂, is below the observed T _g because of rate effects. Enormous periods of time would be required to approach T ₂ experimentally. When discussing the underlying physical cause of the glass transition, Gibbs ³¹ likened the system of polymer molecules to an assembly of tree branches; the polymeric system on contraction and chain stiffening with lowering temperature resembles a compacted brush pile at T ₂ incapable of much further compaction or overall “twig” (i.e., “polymer chain”) motion at lower temperatures. The time effects near T _g were associated with the onset of increasing chain stiffness and degree of compaction as the system approached T ₂. All this may be accepted as a highly reasonable general picture for amorphous polymers, but it leaves the ceramists and metallurgists to devise somewhat different analogies for inorganic and metallic glasses, just as one must do for organic glasses formed by supercooling simple molecules such as glucose and isoamyl bromide.

In glass transition the time effects near T _g follow an activation energy law far different from the customary Arrhenius law. For example, the fluidity ϕ (i.e., the inverse of the viscosity η) for simple liquids at high temperatures follows the Arrhenius form

where E* is the activation energy for viscous flow. However, in the subcooled state in the general vicinity of T _g , an amorphous polymer always exhibits the behavior

where U* is also an activation energy and T _∞ a temperature from 20°C, say, to 50°C below T _g . Thus the fluidity acts near T _g as if it would vanish at the finite temperature T _∞! Put another way, embrittlement sets in rapidly in a rather short range of temperature. Is T _∞ to be identified with the Gibbs-DiMarzio T ₂? It is tempting to say yes, but the truth is that we do not know. The authors are aware of no completely satisfactory derivation of the exp[—U*/ R(T—T _∞)] empirical law. “Free volume” concepts can be used to deduce a form like it, but pressure experiments (where constant volume can be maintained over a range of temperature) make trouble for this approach. This law is probably in the general class of time-dependent cooperative effects. We note also that the semiempirical law that polymer scientists refer to as the Williams-Landel-Ferry (WLF) equation, ³² was known independently by glass technologists as the Fulcher equation and by others. Further elucidation of the glass transition, especially the basis of its related time effects, stands as a challenge to ceramic, metal, and polymer scientists alike.

The following discussion of some current applications of polymers is presented (1) to retire the myth that polymers are not often the material of choice, (2) to show how polymeric materials support other major materials-intensive technologies interactively, (3) to indicate how polymers contribute to human benefit, (4) to demonstrate that progress is based on scientific foundations, and (5) to indicate how polymers work in conjunction with other materials. It is not often realized how pervasive polymers have become in daily life and in science and technology. We are inured to seeing plastic food wrappings and similar applications of polymers but are not often aware of a more complete set of applications.

There are, of course, many current applications of polymers, from contact lenses to containers. Following is a brief discussion of four applications: piezoelectric polymers, polymer precursors for ceramics, photoresists for silicon chip technology, and implants in the human body.

One of the most prominent materials exhibiting piezoelectricity is a crystalline, processible polymer: polyvinylidene fluoride,—(CH₂—CF₂) _n —. It is a simple polymer—“polyethylene” again but with fluorine atoms replacing the protons every other carbon atom. Its three known molecular chain conformations are shown in Figure 15 in space-filling molecular models. The diagrams below the models show the carbon and fluorine atoms (protons omitted for clarity) as viewed down the chain. At each substituted carbon atom there is a large net electric dipole moment due to the highly polar—CF₂—group. These three molecular conformations pack into a total of five known crystallographic phases (polymorphs). Projections on (001) of the two most important of these are shown in Figure 16, i.e., forms I (β) and II (a). Form II crystallizes from the quiescent melt and form I is obtained by stretching form II films at low temperatures (60–150°C). Each polymer chain has a net dipole moment indicated in the figure by an arrow. Clearly, the chain dipole moments cancel in form II, whereas they reinforce in form I. Application of a high electric field (“poling”) reorients the dipoles of the crystal phase preferentially in the direction of the field, and they remain

aligned upon removal of the field. ³³ This interesting form of processing makes the material piezoelectric and apparently also ferroelectric. ^{33 , 34} Since polyvinylidene fluoride is moldable, transducers are readily fabricated.

In a related vein, many polymers show conductivity when they are doped (both n and p). Examples are shown in Figure 17. When these are doped, electrons can move freely along the polyconjugated paths. So far as we know, commercial applications are yet to be achieved but are surely to be expected. The activity in this field is considerable, and a conscientious review of it would be lengthy.

The detailed discussion concerning molecular conformations and crystal structures of polyvinylidene fluoride was included in part to illustrate that

there is an enormous body of information on the crystal structure of polymers. Hundreds of crystal structures are known. ³⁵ Here again polymer science owes a debt, this time to those who developed the principles of x-ray crystallography. The polymer scientists have since added nuances of their own to deal more effectively with special problems related to polymers.

It is often difficult to produce molded objects (complex shapes) from ceramics because of their hardness and the extremely high temperatures that are required. A sometimes easier approach uses metal-organic polymer precursors that are processible and that are then converted to a ceramic body of the desired shape by appropriate pyrolysis (firing) techniques. Consequently, polymeric ceramic precursors are a potential source of coatings (for corrosion resistance, abrasion resistance, electrical insulation, and optical transmission/reflectance), of powders and sintering aids in the preparation of special-purpose bulk ceramics, and of fibers for high-temperature, high-strength composites. Again, the international character of research and technology in this field is evident.

The first example (Figure 18) is a process developed by the Japanese company, Nippon Carbon, to make the fiber called Nicalon^®. The dimethylpolysilane polymer at the left is converted to a polycarbosilane. Both polymers are tractable and can be melt-spun to make fibers in a manner roughly analogous to the spinning of nylon. After curing, pyrolysis of the polycarbosilane fiber yields an extremely good silicon carbide (SiC) fiber (with the side products indicated). Objects molded or fabricated from the polymer precursor will, after pyrolysis, be silicon carbide ceramics. Note that SiC itself is difficult to shape. In this example, the SiC fiber was obtained by combining polymer technology and knowledge of how polymers degrade under heat. Study of Nicalon SiC fibers was conducted by the Celanese

Research Company and by Dow Corning under a contract with the Defense Advanced Research Projects Agency. Fracture surfaces of a 10-µm Nicalon fiber are shown in Figure 19 (a matched pair of ends). Each picture shows a classical Griffiths fracture surface with initiating flaw, mirror (smooth area around the flaw through which the crack traveled in a controlled fashion), and the rougher surface indicating catastrophic crack propagation. The cleaner and purer the fiber, the stronger and more reliable is the material. With appropriate care, the heterogeneities that initiate crack propagation can be minimized and excellent-strength SiC fibers formed.

The Dow Corning precursor is a polydisilylazane, a mixed polymer with monomer segment concentrations. After pyrolysis, one has a mixed-composition silicon carbide-silicon nitride (SiCN) fiber about 10 µm in diameter. By changing the polymer chemistry, cleverly synthesizing this molecule, spinning it as a tractable polymer, and pyrolyzing it, Dow Corning has made superior materials with tensile strengths of approximately 300 kpsi. This is an important area of emerging technology.

Briefly, virtually every silicon (also gallium arsenide) chip now in use was made using positive or negative photoresists that are polymeric in character. The process is shown in Figure 20. The photoresists control the areas of the chip that are made into conductors or insulators in subsequent evaporation operations; they are absent in the final product. Silicon chip technology would not have been developed without a major contribution of polymer science (much of it chemistry) and technology. This is another example where polymer science and technology assist other materials efforts. The outstanding computer and communication industries in the United States owe much to polymeric photoresists.

Figure 21 is a picture of a hip implant, a good solution to a problem through the intelligent use of a combination of different classes of materials. Currently, two different polymers and a metal are used. The hip cup is made of ultrahigh-molecular-weight polyethylene (again!), which is put in the pelvis and virtually never fails. The shaft is vitalium metal. It is difficult to

find a noncorrosive metal that is acceptable to the human body, although the vitalium generally works out well. The vitalium shaft is cemented to the femur with methyl methacrylate monomer, which is then polymerized in situ during the implant operation on the patient. The problem of controlling the polymerization reaction under these conditions can be imagined. Too rapid a polymerization means too much heat generation, which can be injurious, and too slow a polymerization can fail to give sufficient physical strength before the operation is complete. But the use of the methyl methacrylate, derived in part from dental technology, does indeed work. This method of attachment of the metal to the femur introduces a possible mode of failure—the metal shaft working loose in the polymethyl methacrylate binder. Improvements are undoubtedly possible, but these hip implants are a great boon to those who need them.

One measure of the importance of this application of polymer science and technology is the 130,000 hip implants per year using polyethylene. Because many of these are put into younger persons, it is appropriate to be concerned with failure analysis. The high molecular weight of this material, while subduing the crystallinity somewhat, goes a long way toward increasing impact strength and wear resistance. With its ultrahigh molecular weight and good toughness, this inert and reasonably tractable material has a very important application. The scientific foundations of such implants are impressive: studies of polymer chemical kinetics, polymer degradation, mechanical strength and wear, metal stress corrosion, and clinical uses all had to be done. This is a good example of the combined use of metals and polymers for human benefit.

The range of physical properties achievable by intelligent use of chemistry and an understanding of property-structure relationships is considerable, as we have seen. One more example will serve to elucidate this point further. By the appropriate choice of “ring opening” monomers, Bailey et al. ³⁶ have shown that it is possible to make practical polymers that either contract (which is the usual case) or expand upon polymerization. Moreover, in a well-defined middle ground of composition, neither expansion nor contraction occurs. Such behavior is interesting to contemplate, and projected uses include dental applications where low-to-moderate expansion is desirable. Also to be considered here is the possibility of potting delicate electronic components to protect them from mechanical damage and hostile environments.

In this section some expectations for future uses of polymers are discussed. Undoubtedly we have overlooked many possibilities, but the field is rich enough to permit many surprises. There will be improved composite structures that are readily manufacturable and of predictable lifetime. There will be improved blends (polymer alloys) that are also readily manufacturable and have a predictable lifetime. Much has been accomplished on blends, but a lot remains to be done. Lightweight batteries and high-quality aspheric lenses will become available. Clearly, camera manufacturers have already made advances here for their current models of small cameras. Corrosion resistance will be improved. The possibility of self-assembling polymeric chips may be farfetched but cannot be ignored. Practical polymers from direct biological conversion of inexpensive feedstocks (e.g., waste) are clearly possible—polyhydroxybutyrate, which in many respects rivals polyethylene in its physical properties, is currently obtained in this manner. Escherichia coli and some other bacteria can manufacture good, very pure polymers of known molecular weight and molecular weight distribution. This, of course, is what an important part of biotechnology is all about.

Two extremely important subjects illustrate succinctly the current technology, current knowledge, and directions of future advances. First, there are blends, which may be either block copolymers (two or more chemically different polymer chains connected together by a chemical bond) or polymer alloys (two or more chemically different polymers that are mechanically mixed).

The manner of blending the two different polymers strongly affects the resultant properties—for example, the important property of impact strength (as mentioned earlier). Figure 22 shows this for blends of thermoplastic polystyrene (S) and polybutadiene (B). With mechanical mixtures (simple blends), impact strengths show little or no improvement. However, if the proper kind of block copolymer is made with chemical linkages between the two types of chains, superior impact strengths are achieved. The different types of copolymer that can be made are listed in Table 2 together with approximate tensile strengths. One can see that most of the possible copolymers offer little improvement in properties. However, with one of the possible triblock copolymers (S-B-S) there is a marked increase in tensile strength as well as in impact strength (Figure 22). The reason for this behavior is shown in Figure 23, in which the morphologies of S-B diblock and S-B-S and B-S-B triblock polymers are represented. In this figure, the butadiene (B) portion is shown by the solid lines and the styrene (S) portion by the broken lines. In each case there is an aggregation of one species (the styrene). Only in the case of the S-B-S block copolymer, however, does this aggregation create a three-dimensional network, which in turn leads to the improved toughness and tensile strength. This is another example of the effect of molecular architecture on properties, in this case on the important property of strength.

As already indicated (Figure 23), phase separation of the polymeric species occurs. The phase-separated blend is in effect a self-assembled composite but of a different origin than that in the semicrystalline polymers mentioned earlier. This phase separation is of a most unusual type because one part of a given giant molecule is in one phase and the other part is in a second phase. The phase domains are thus of molecular dimensions, typically 10 to 30 nm in diameter. This is a curious situation, but thermodynamics is not violated.

Thus, the whole question of phase diagrams becomes as important for polymer blends as it is for metals and ceramics. It is also necessary to know how fast phase separation occurs and whether it is spinodal decomposition or ordinary phase separation. The nature and properties of the interfaces are important.

There is greatly increasing emphasis in both fundamental and applied work in this field. Many commercial applications have arisen with gratifying improvements in properties. Much basic work is going on in studies involving phase diagrams, dynamic mechanical behavior, and attempts to understand impact strengths and the size and nature of interfaces. Progress is already highly significant, but much remains to be done. Some of the needs for new knowledge in the field of blends are

The second area in which much is being done is composites. Nature invented composites—wood (cellulose and lignin) and bone (the polymer collagen and the mineral hydroxyapatite) are specific examples. A composite can and often does have much more desirable properties than do the individual “pure” or “virgin” materials from which it was made. Man-made composites have also been successful, as in the case of the “rubber” tire, which in its most common modern form is a composite of vulcanized rubber (the synthetic or natural polymer), carbon filler, and steel or polymeric fibers. One reason for interest in other man-made composites is indicated in Figure 24, which compares specific strengths (tensile strength/density). The high strength-to-weight ratio of composites is more favorable than their ratio of strength to size. The high strengths of the aramid Kevlar and graphite composites justify commercial interest in them. Glass composites combine a slight sacrifice in properties with a significant drop in cost. Current commercial aircraft use substantial amounts of nonstructural composites and about 37 percent by weight of composites in primary structure. Composites are an absolutely essential component in modern military aircraft. The current in-service airplane contains alloy steel in the engines, aluminum over the body, some titanium, and various types of composites. Current commercial aircraft design

contemplates an increased use of composites; for future subsonic (not supersonic) aircraft, the potential composite use in primary structure exceeds 50 percent by weight of the aircraft. The net result will be a considerable weight saving, with a concomitant increase in fuel efficiency, as well as highly satisfactory durability, strength, and corrosion resistance. For many aircraft structural members, composites have become the materials of choice.

In a related field, the body of the popular and impressive Pontiac Fiero automobile is made of composites. Lest one believe that the use of polymers for automobile parts and bodies is entirely new, one of the authors had in his possession a photograph taken in 1942 showing Henry Ford wielding an axe on a plastic trunk door made from soybeans (there is no photograph of the sequel). The science of polymers was in its infancy in those days. The idea, strongly espoused by Staudinger, ³⁷ that polymers were giant molecules had not yet taken full hold. (It now seems almost incredible that Staudinger, rightly deemed the father of polymer science, had an extremely difficult time convincing many of the scientists of his day that giant molecules existed. Some of the resistance undoubtedly arose, apart from the natural conservatism of most scientists, from the then strong presence of influential scientists who refused to believe even in atoms, because thermodynamics, mechanics, and optics had been developed so successfully without them.) The current scientific basis is much improved, and more rapid progress can now be made. The increased use of composites in automobile manufacture is a virtual certainty.

Although not emphasized up to this point, most of the current polymer formulations employed in composites in aircraft and automobiles are three-dimensional networks, the so-called epoxy compounds being typical. These networks are similar to those shown in Figure 23, except that the cross-links involve chemical bonds. These materials (“thermosets”) are converted from the monomer-fiber-filler mixture (which is like a slurry) by the introduction of a chemical catalyst and application of heat in a curing cycle. The result is a tough composite consisting of high-molecular-weight polymer bonded to the fiber and the filler. Voids can be a problem. Because of the time required for the procedure, thought is currently being given to using thermoplastic polymers (i.e., polymers that are softened by heat and thereby rendered rapidly moldable) in composites. Such thermoplastic polymers might be of either the crystallizable or amorphous type; if crystalline, the crystals themselves can act as cross-links between chains. Rapid manufacture of a finished part having uniform high quality and predictable properties is the overall goal.

Current problems and issues for further study concerning composites certainly include their processing and manufacturability and the manner in which they fail. Damage to an object made of metal (such as a car) results in a visible dent, and repair is relatively simple. A composite can be damaged badly and frayed on the inside but show little sign of it on the outside. ³⁸ Repair tends to be difficult. The detection of flaws with nondestructive evaluation is a major issue in composites. A basic scientific understanding of the mechanism of failure and of the lifetime durability of composites is a high-priority subject.

In this regard, a research briefing ³⁹ prepared at the request of the Office of Science and Technology Policy and the National Science Foundation (under the joint chairmanship of D.W.McCall of AT&T Bell Laboratories and R.Pariser of E.I.du Pont de Nemours & Co.) recommended directions for research. Among other concerns, it is necessary to obtain a better understanding of the relationships between molecular structure and physical properties of fibers and how these relationships can be translated into the behavior of a fiber in a composite. Since composites are subject to long-term cyclical loading, an understanding of the fatigue behavior of fibers under such conditions is necessary. Other key issues for further study are the fundamentals of the fiber-polymer interface as related, for example, to adhesion and load bearing; methods for joining or fastening composites to like or different materials; and control of creep under load (desired also for metals). The best way to resolve these issues is not clear. Efforts have been made, with some success, but there remains a great challenge. A common current practice is to pick a specific system and then explore that system in depth. But can broad generalizations be made? The answer, it is hoped, is “yes,” but the information must be sought out with vigor.

The issues involved in the processing of polymer composites include the chemistry of the composite itself (what materials are chosen), the chemical interactions at the interfaces with the filler and the fiber, the flow and other problems associated with manufacture, and the nature of the structure finally achieved. Fundamental scientific and applied technological attacks on all aspects are required. It should come as no surprise that some of the flow properties encountered in the “slurries” that make up pre-composites before molding or curing are related to concepts developed by soil scientists, who have studied the behavior of moist earth under stress for a long time.

For both blends and composites there is a clear need to understand chemical, physical, processing, and lifetime behavior. Approaches must be interdisciplinary, based both on theory and on experiments in organic chemistry, physical chemistry, rheology, and solid-state physics. They must include the fields of metals, polymers, ceramics, interfaces, and nondestructive evaluation. This breadth of endeavor itself poses a large and difficult challenge to scientists and engineers. Cooperation and interaction among industrial, academic, nonprofit, and government organizations will be important. Japanese and European competition is already evident, but the future holds sufficient promise to make the effort well worthwhile.

The universities have made a significant response to the need for education and research in composites. Some of the institutions involved are the University of Massachusetts, Washington University, the University of Delaware, Virginia Polytechnic Institute, and the Case Western Reserve-University of Akron dyad. This list is, of course, not exhaustive. Certain universities, such as Michigan State University, Michigan Technological University, and the University of California, Santa Barbara, have serious intentions in the field. Doubtless, there will be more.

The authors’ organization, Michigan Molecular Institute, plans to cooperate with universities on both education and research efforts in the areas of blends and composites. We believe that a strong materials science base is an absolute prerequisite for such an activity. Industrial research laboratories are also highly active in the area—the aerospace industries heading the list—and it is clear that great interest and activity have also developed in the automotive company laboratories.

Governmental involvement is not lacking either: the highly significant programs at Wright-Patterson Air Force Base and by the National Aeronautics and Space Administration are well known, and recently the National Bureau of Standards developed a response. The National Science Foundation is fully aware of the issues involved and has provided significant support to some of the universities mentioned.

The final topic of this review concerns aspects of the newer theories of polymer science currently stimulating the field: reptation, phase transitions, and some unusual behavior and generalizations concerning polymer melts. Reptation has already been mentioned and will be discussed again shortly.

The “trajectory” of a flexible polymer molecule in the molten state is not unlike the path of an inebriated bumblebee, except that, because the chain atoms occupy space, the chain cannot cross itself. Consequently, the chain is slightly expanded and is called the excluded-volume coil (Figure 25, left). If there were no self-exclusion, the chain would be somewhat smaller (Figure 25, center); certain solvents (called theta solvents) permit the chain to act like this. If the pH or the thermodynamic driving force of the solvent is changed

(i.e., the “goodness” of the solvent is changed), a collapsed coil can appear abruptly (Figure 25, right). In other words, there is a true phase transition within a single molecule! This possibility has been of great interest to theoreticians. This “collapse transition” was originally predicted by Stockmayer, ⁴⁰ Monte Carlo studies suggested it, ⁴¹ and the newer renormalization group and scaling theories ²⁶ have been able to deal with it effectively. The problem is similar—practically identical mathematically in one limit—to the magnetic spin problem in ferromagnets. ⁴² This for a linear, flexible polymer! Recently, the collapse was observed experimentally. ⁴³ Figure 26 summarizes the theories and observations. The mean-square radius of gyration, is related to the length, N, of a polymer chain raised to some power, γ, where γ is a measure of the solvent power. The schematic shows an abrupt change in size with solvent character, with the transition actually occurring over a few tenths of a degree. In other words, polymer chains in solution behave differently than do low-molecular-weight compounds. One fascinating side effect of all this is that articles relating to the aforementioned phase transition in polymers are now occasionally found in the Physical Review, a journal that has in the past carried very few articles on polymers. It must be pointed out, however, that neither the American Chemical Society nor the American Physical Society ignores the topic of polymers; on the contrary, both strongly support active divisions for polymer science.

Polymer melts also show unusual behavior when compared with normal fluids. For example, an ordinary liquid pumped out of a tube will exhibit the profile left-hand part of Figure 27. ⁴⁴ A polymer liquid (without confinement) pumped out of a tube swells up on exit, as shown on the right-hand

side of Figure 27; i.e., it exhibits “memory.” In many systems, such as metals, the deviation from perfect elasticity is usually small. In polymeric systems, by contrast, mechanical behavior is frequently dominated by such viscoelastic behavior with its pervasive memory effects. Bird and Curtiss ⁴⁴ have illustrated many other differences in the behavior of simple and polymeric liquids. (We have already mentioned one—namely, the behavior of the fluidity ϕ with temperature.) The unusual mechanical behavior of polymer melts is governed in general by nonlinear viscoelastic theory. To understand manufacturability and processing in polymers better, one must understand not only such simple behavior but also many complex phenomena, of which the example of Figure 27 is but a premonitory hint. One wants a molecular view of these effects that could be reduced to practice.

However, the polymer engineers and phenomenologists prefer to think in terms of continuum equations, called constitutive equations. An example of a generalized simple linear constitutive equation is

where γ₂₁ is the shear strain at time t′ relative to that at time t. In a sense this approach is similar to thermodynamics in that there is no presence of the molecule in such equations—i.e., no molecular parameters. This equation implies that the shear stress, σ₁₂, is obtained by an integration of the product of some kind of memory function, m, and the shear strain.

This, in a general sense, is the sort of formulation required to understand phenomena such as that illustrated in Figure 27. There is no lack of inspired constitutive-based equations to deal with such situations; an example is the useful Bernstein-Kearsley-Zapas (BKZ) theory. ⁴⁵ But, as we have said, we would like to know about the role of the molecules.

The reptation model (which is molecular in character) discussed earlier is thought to be applicable here and is shown with the subject molecule and its confining tube in an isolated state in Figure 28. ²⁶ Recently, Doi and Edwards, ²⁷ de Gennes, ⁴⁶ Curtiss and Bird, ⁴⁴ and Graessley ⁴⁷ have begun to modify the reptation model by, for example, letting the tube diffuse around on its end. On the basis of this type of molecular model, in one simplistic form, they have been able to derive an elementary constitutive equation from molecular considerations. It is hoped that such an approach, if it is not mathematically intractable, will lead to a much-improved molecular picture of what is happening when polymers are processed. The ultimate goal is to relate chain (i.e., molecular) properties to stress-strain-time relations in a polymer melt so that processing may be understood in basic terms. A special need is perceived for what may indeed be a new language that will allow scientists and engineers (and project managers) to communicate better concerning polymer properties as they relate to practice. Quoting the modulus is no longer enough, if indeed it ever was. Neither is a computer simulation that slurs over material properties. The need is great enough for polymers in conventional use and surely reaches its zenith when composites of polymers with metals or ceramic bodies are considered.

The field of polymer science is displaying a tremendous vitality and energy, coupled with high-quality science. It now involves basic scientists from seemingly distant fields: theoretical physics (phase transitions); solid-state physics (piezoelectric, conductive, and semiconductive polymers); statistical mechanics; quantum mechanics; continuum mechanics; biophysics; and biochemistry. In addition, rheology and viscoelasticity are its special province. There is plenty of scope for intellectual curiosity and creativity, with many unsolved problems even in the “conventional” parts of the field. In these, as well as in some of the new directions we have noted, there is a high probability that the science will lead to practical results and benefits. However, progress will depend in considerable degree on broad and fundamental knowledge and training in materials science as a whole; as we have illustrated repeatedly, a single type of material no longer stands in total isolation from the others, and the basic disciplines are still fundamental to every aspect of materials science.

In preparing this review the efforts of numerous others were invaluable, specifically C.M.Guttman, G.T.Davis, L.Smith, and D.Huntston of NBS; M.Jaffe and R.M.Mininni of Celanese Corporation (in cooperation with DARPA and Dow Corning); K.Bowen of MIT; DARPA; R.E.Hefner, consultant to Michigan Molecular Institute; D.J.Meier of MMI; D.W. McCall of AT&T Bell Laboratories; J.T.Quinlivan of Boeing Company; R.S.Porter of the University of Massachusetts; and R.K.Eby of the Johns Hopkins University.